An unfortunate result of technological advances in image capture and reproduction is illegal copying and distribution of image content, in violation of copyright. One solution for counteracting illegal copying activity is the use of image watermarking as a forensic tool. Sophisticated watermarking techniques enable identifying information to be encoded within an image. A watermark can be embedded in the image beneath the threshold of visibility to a viewer, yet be detectable under image scanning and analysis. As just a few examples: U.S. Pat. No. 6,239,818 (Yoda), discloses embedding a pattern in a color print and adjusting cyan, magenta, yellow, black (CMYK) values such that the embedded data matches the color of the surround when viewed under a standard illuminant; commonly assigned U.S. Pat. No. 5,752,152 (Gasper et al.), discloses a pattern of microdots, less than 300 μm in diameter, for marking a photographic print that is subject to copyright.
Illegal copying is a particular concern to motion picture studios and distributors, representing a noticeable source of lost revenue. Watermarking of motion picture images would enable the source of an illegal copy to be tracked and would thus provide a deterrent to this activity. Watermarking techniques for still images and prints, however, may not be well-suited to motion picture film media. An encoded pattern that might not be easily visible within the single image of a print could become visible and annoying if it appears in a sequence of image frames. Moreover, a motion picture watermark must be detectable from a copy, such as a videotape copy, that is typically captured in a temporal sequence that varies from the motion picture projection rate and with varying image resolution, lighting, and filtering. For these and related reasons, motion picture watermarking typically requires a special set of techniques beyond those normally applied for still images.
A number of watermarking methods for motion images have been described in prior art patents and technical literature. Included are methods that apply a spatial-domain or frequency-domain watermark. In either approach, many techniques make use of a pseudo-random noise (PN) sequence in the watermark generation and extraction processes. The PN sequence serves as a carrier signal, which is modulated by the original message data, resulting in dispersed message data (that is, the watermark) that is distributed across a number of pixels in the image. A secret key (termed a “seed value”) is commonly used in generating the PN sequence, and knowledge of this key is required to extract the watermark and the associated original message data.
Among prior art patents that address watermarking methods for motion picture image content are U.S. Pat. No. 5,809,139, issued Sep. 15, 1998 to Girod et al., entitled “Watermarking Method and Apparatus for Compressed Digital Video”; U.S. Pat. No. 5,901,178, issued May 4, 1999 to Lee et al., entitled “Post-Compression Hidden Data Transport for Video”; and U.S. Pat. No. 5,991,426, issued Nov. 23, 1999 to Cox et al., entitled “Field-Based Watermark Insertion and Detection”. However, the methods disclosed in these patents can be applied only to a digital video data stream and are not directly applicable for watermarking motion picture film.
U.S. Pat. No. 6,026,193, issued Feb. 15, 2000 to Rhoads, entitled “Video Steganography”, discloses the basic concept of using multiple watermarked frames from an image sequence to extract the watermark with a higher degree of confidence than would be obtained with only a single frame. U.S. Pat. No. 6,449,379 issued Sep. 10, 2002 to Rhoads, entitled “Video steganography methods avoiding introduction of fixed pattern noise” proposes an improvement to this scheme by changing the PN carrier from frame to frame, for example.
Another approach to applying a watermark to an image sequence is to use a three-dimensional watermark pattern. An example of such a method can be found in a paper by J. Lubin et al, “Robust, content-dependent, high-fidelity watermark for tracking in digital cinema,” in Security and Watermarking of Multimedia Contents V, Proc. SPIE, Vol. 5020, Jan. 24, 2003. This paper discusses a method for embedding, into successive image frames, a watermark containing low frequency content in both the spatial and temporal dimensions. The method described by Lubin et al. may provide a temporally distributed watermark that is relatively robust. However, this method requires temporal synchronization in order to recover or decode the watermark. That is, some mechanism must be provided that allows indexing of each image frame with a reference frame; a sampling of successive image frames must include this reference in order to allow synchronization of watermarked frames and subsequent decoding. Another limitation is that knowledge of the image content is required for embedding a three-dimensional watermark using this scheme.
The prior art methods cited above provide some amount of watermarking capability, but are not well-suited for printing watermarks onto motion picture film media. This is because each of these methods requires that frame boundaries for each image be known prior to printing the watermark onto the motion picture film medium. That is, the precise location of image frames on the motion picture medium must be known.
For photosensitive media in general, it is known that a watermark encoding can be added to the image frame at the time of printing. However, it is also possible to expose a watermark at other times during processing of the photosensitive medium. For example, as is disclosed in U.S. Patent Application 2003/0012569 entitled “Pre-Exposure of Emulsion Media with a Steganographic Pattern” by Lowe et al., a latent image can be exposed onto the “raw” photosensitive medium itself, at the time of manufacture. Then, when the medium is exposed to form the image, the image frame is effectively overlaid onto the watermark pattern. Such a method is also disclosed in U.S. Pat. No. 6,438,231, entitled “Emulsion Film Media Employing Steganography” to Rhoads. The Rhoads '231 patent discloses this type of pre-exposure of the watermark onto the film emulsion within the frame area of negative film, for example.
It can be appreciated that watermark pre-exposure would have advantages for marking motion picture film at the time of manufacture or prior to exposure with image content. A length of motion picture film could be pre-exposed with unique identifying information, encoded in latent fashion, that could be used for forensic tracking of an illegal copy made from this same length of film. However, prior art watermarking techniques proposed for photosensitive media in general fall short of what is needed for motion picture watermarking. In particular, prior art techniques are not well-adapted for applying a watermark pattern during film manufacture or at any other time prior to exposure of the film with image content. These prior art solutions prove unsatisfactory due, in large part, to these practical considerations:                (i) how motion picture film is imaged and projected in practice; and,        (ii) how the motion picture film surface area or “real-estate” is employed.        
Referring to FIG. 1, there is shown a plan view of a typical motion picture print film 10 that is used commercially. A first problem ((i) above) relates to the placement of frames 12 along the length direction L. Frames 12 are dimensioned and spaced with a frame pitch F according to well-established standards followed, for each film type, throughout the motion picture industry. In addition, perforations 14 are made in the film, sized and spaced apart according to rigid standards. For 35 mm film, for example, manufacturers comply with the Society of Motion Picture and Television Engineers (SMPTE) Standard ANSI/SMPTE 139-1996 entitled “SMPTE Standard for Motion-Picture Film (35-mm) Perforated KS”. Perforations 14 are formed at the time of film manufacture. Frames 12, however, are not formed until printing at the print lab. Thus, the exact locations of frames 12 along print film 10 are not yet defined at the time of manufacture. The starting position of each frame 12 may be known relative to an index perforation 14; however, it is not known which perforation 14 is used as a reference index until the photosensitive medium is exposed. Even though frames 12 correlate spatially with perforations 14, so that each frame 12 plus interframe space 16 corresponds to an exact number of perforations (typically 4 perforations per frame), the film manufacturer cannot know beforehand where each frame 12 will lie. Thus, unlike the frame-by-frame placement suggested for the broad range of photosensitive media in the Rhoads '231 disclosure, any practical exposure watermarking scheme for motion picture film must apply a contiguous pattern along the full length of film. This means that the watermark pattern will be exposed onto both image frames 12 and interframe spacing 16 areas.
This first concern, then, relates to dimensional characteristics of the watermark. Tiling, in which multiple versions of a watermark are repeated contiguously within the image frame, has been widely recognized as a useful method for encoding the watermark. For example, commonly assigned U.S. Pat. No. 6,044,156 entitled “Method for Generating an Improved Carrier for use in an Image Data Embedding Application” to Honsinger et al. discloses an image watermark scheme using one or more tiles. FIG. 2 shows an example in which multiple watermark tiles 20 are contiguously arranged within image frame 12. (For illustration, only a portion of image frame 12 is shown covered in FIG. 2; in actual practice, the complete area of image frame 12 would have watermark tiles 20.) Watermark tiles 20 can be arranged contiguously in both length L and width W dimensions, extending widthwise between perforations 14 or even between opposite edges of the medium.
Regardless of watermark tile 20 dimensions, the watermark can be detected only when it occurs within image frame 12. However, unless watermark tile 20 is positioned length-wise at the same relative position within successive frames 12, the watermark will effectively “walk” up the displayed motion picture frame when the motion picture image is displayed. While this beat-frequency effect might not be detectable over portions of an image sequence that are themselves visually busy, this “walking” effect could be noticeable and visually objectionable at some frequencies and under some imaging conditions. More importantly, changing the vertical position of watermark tiles 20 from one frame 12 to the next complicates the task of detecting the watermark. This first consideration, then, directly affects the robustness of the watermark. The disclosure of commonly assigned copending application “Method and Apparatus for Watermarking Film” to Roddy et al., U.S. Pat. No. 6,882,356, cited above, recognizes this vertical alignment problem and proposes, as a corrective strategy, sizing the watermark tile height to span a single pitch perforation. While this provides a workable solution in view of problem (i) noted above, it constrains the tile height dimensions to a single perforation pitch, where perforation pitch can be defined as the center-to-center distance between perforations. It can be appreciated that a more flexible solution would eliminate this constraint and allow more flexible sizing of watermark tile height, based on film type, for example.
The second problem ((ii) above) relates to the use of the width W of print film 10. As FIG. 1 shows, the real-estate of the surface of print film 10 is occupied not only by the sequence of frames 12, but also by various audio tracks. An analog sound track 18 is printed between the side edge of frames 12 and perforations 14. A DTS (Digital Theater Systems) soundtrack 26 is encoded between frames 12 and analog sound track 18. A Dolby digital sound track 22 uses areas interspersed between perforations 14, repeated on both sides. Another digital sound track 24, conventionally the standard SDDS (Sony Dynamic Digital Sound) track is encoded along edges of print film 10. Digital sound tracks 22 and 24 are typically duplicated on both sides of print film 10 as indicated by digital sound tracks 22′ and 24′. For considerations of watermark application, it is significant to observe that analog sound track 18 and digital sound tracks 22, 24, and 26 are encoded onto print film 10 using exposure to light, in much the same way as frames 12 are exposed. For this reason, any imperfection in imaging quality of print film 10 may also impact audio quality. Film grain, dust, surface imperfections, and other imaging anomalies not only degrade image quality, but may also have an impact on audio quality. For this reason, a watermark that is applied along edges of print film 10 used by audio tracks can potentially degrade the encoded audio signal.
A further complication, related to problem (ii) as given above, is that there is no pre-determined widthwise placement of frames 12 and analog sound track 18 and DTS sound track 26 for unexposed film. As the film is shipped from the manufacturer, one orientation is more likely than its opposite; however, either negative or print film may be rewound before being exposed. Therefore, once print film 10 is manufactured, it cannot be determined in which direction print film 10 will actually be exposed. Thus, it is not known at the time of manufacture whether analog sound track 18 and DTS sound track 26 run along the line of perforations 14 nearest one edge of print film 10 or the other. As is observable in the plan view of FIG. 1, frames 12 are skewed to one side of print film 10 relative to width W, rather than being centered, to accommodate audio sound track 18 and DTS sound track 26.
A practical watermark exposure scheme, then, must address the problems of uncertain placement of frames 12 relative to length L and width W, which directly affects robustness and straightforward detection, and of the need for encoding analog and digital sound tracks 18, 22, 24, and 26. That is, the watermarking scheme that is used must address the problems posed in considerations (i) and (ii) given above. Conventional approaches, such as simply applying a watermark pattern from one edge of film 10 to the other, could yield unsatisfactory results, degrading image quality, degrading audio quality, and compromising the robustness needed. Some improvement over conventional approaches is needed for providing watermark exposure, particularly for motion picture film media, that offers a good measure of robustness without introducing problems related to image and audio quality.